Betsy Jane Becker

Physical activity and psychological well-being in advanced age: a meta-analysis of intervention studies.

A meta-analysis examined data from 36 studies linking physical activity to well-being in older adults without clinical disorders. The weighted mean-change effect size for treatment groups (d(C). = 0.24) was almost 3 times the mean for control groups (d(C). = 0.09). Aerobic training was most beneficial (d(C). = 0.29), and moderate intensity activity was the most beneficial activity level (d(C). = 0.34). Longer exercise duration was less beneficial for several types of well-being, though findings are inconclusive. Physical activity had the strongest effects on self-efficacy (d(C).= 0.38), and improvements in cardiovascular status, strength, and functional capacity were linked to well-being improvement overall. Social-cognitive theory is used to explain the effect of physical activity on well-being.

How meta-analysis increases statistical power.

One of the most frequently cited reasons for conducting a meta-analysis is the increase in statistical power that it affords a reviewer. This article demonstrates that fixed-effects meta-analysis increases statistical power by reducing the standard error of the weighted average effect size (T.) and, in so doing, shrinks the confidence interval around T.. Small confidence intervals make it more likely for reviewers to detect nonzero population effects, thereby increasing statistical power. Smaller confidence intervals also represent increased precision of the estimated population effect size. Computational examples are provided for 3 effect-size indices: d (standardized mean difference), Pearsons r, and odds ratios. Random-effects meta-analyses also may show increased statistical power and a smaller standard error of the weighted average effect size. However, the authors demonstrate that increasing the number of studies in a random-effects meta-analysis does not always increase statistical power.

The Synthesis of Regression Slopes in Meta-Analysis.

Research on methods of meta-analysis (the synthesis of related study results) has dealt with many simple study indices, but less attention has been paid to the issue of summarizing regression slopes. In part this is because of the many complications that arise when real sets of regression models are accumulated. We outline the complexities involved in synthesizing slopes, describe existing methods of analysis and present a multivariate generalized least squares approach to the synthesis of regression slopes.

Multivariate Meta-analysis

Publisher Summary This chapter reviews meta-analytic methods for synthesis of multivariate data. Meta-analysis, or research synthesis, provides a way to examine results accumulated from a series of related studies, through statistical analyses of those results. The approach to multivariate meta-analysis presented here can be applied regardless of the form of the effect of interest. In general, the goals of a multivariate meta-analysis are the same as those of univariate syntheses: to estimate magnitudes of effect across studies, and to examine variation in patterns of outcomes. Moreover, reviewers have a wide range of options available when faced with multivariate data in meta-analysis. The specific set of analyses a reviewer selects depends on a number of factors. For example, the reviewer must consider the structure of the data in the review. If the multivariate data are a small part of the evidence in the review, it may make sense to use a simple approach, such as dropping or combining outcomes, or using sensitivity analyses to evaluate the impact of the dependence on the results of the review.

Using Results From Replicated Studies to Estimate Linear Models

This article outlines analyses for the results of a series of studies examining intercorrelations among a set of p + 1 variables. A test of whether a common population correlation matrix underlies the set of empirical results is given. Methods are presented for estimating either a pooled or average correlation matrix, depending on whether the studies appear to arise from a single population. A random effects model provides the basis for estimation and testing when the series of correlation matrices may not share a common population matrix. Finally, I show how a pooled correlation matrix (or average matrix) can be used to estimate the standardized coefficients of a regression model for variables measured in the series of studies. Data from a synthesis of relationships among mathematical, verbal, and spatial ability measures illustrate the procedures.

Coaching for the Scholastic Aptitude Test: Further Synthesis and Appraisal

Analyses of results of studies documented in 23 reports on coaching for the Scholastic Aptitude Test (SAT) indicated that, on average, coaching can help increase SAT scores. However, considerable variability in results for 48 studies reflected the fact that not all coaching is necessarily effective and that all studies of coaching do not provide similar views of coaching’s effectiveness. Results of published and unpublished studies were analyzed separately. Characteristics related to the magnitudes of coaching effects included date of publication of the study, whether the study was sponsored by the Educational Testing Service and used as a comparison group, whether instruction included test practice and attention to test-taking skills, and whether homework was assigned to students. Coaching effects were stronger for the SAT Mathematical subtest. Published comparison studies gave consistent results with coached groups exceeding controls by 0.09 standard deviations on SAT-V and 0.16 on SAT-M. Studies of coaching were found to be rather poorly reported and designed without much attention to the issues discussed in reviews of the coaching literature.

The efficacy of medical abortion: a meta-analysis

Multiple clinical studies demonstrate the efficacy of medical abortion with mifepristone or methotrexate followed by a prostaglandin analogue. However, assessing predictors of success, including regimen, is difficult because of regimen variability and a lack of direct comparisons. This meta-analysis estimates rates of primary clinical outcomes of medical abortion (successful abortion, incomplete abortion, and viable pregnancy) and compares them by regimen and gestational age. We identified 54 studies published from 1991 to 1998 using mifepristone with misoprostol (18), mifepristone with other prostaglandin analogues (23), and methotrexate with misoprostol (13). Data abstracted from studies included regimen details and clinical outcomes by gestational age. We found that efficacy decreases with increasing gestational age (p<0.001), and differences by regimen are not statistically significant except at gestational age > or =57 days. For gestations < or =49 days, mean rates of complete abortion were 94-96%, incomplete abortion 2-4%, and ongoing (viable) pregnancy 1-3%. For gestations of 50-56 days, the mean rate of complete abortion was 91% (same for all regimens), incomplete abortion 5-8%, and ongoing pregnancy 3-5%. For > or =57 days, success was lower for mifepristone/misoprostol (85%, 95% confidence interval 78-91%) than for mifepristone/other prostaglandin analogues 95% (CI 91-98%, p = 0.006). For mifepristone/misoprostol, using > or =2 prostaglandin analogue doses seems to be better than a single dose for certain outcomes and gestational ages. We conclude that both mifepristone and methotrexate, when administered with misoprostol, have high levels of success at < or =49 days gestation but may have lower efficacy at longer gestation.

A Look at the Literature (and Other Resources) on Teaching Statistics.

Statistics instructors and others interested in the teaching of statistics will find many print and nonprint resources on this topic. The print literature on the teaching of statistics is largely anecdotal and comprises mainly recommendations for instruction based on the experiences and intuitions of individual instructors. Less than 30% of the print literature reports the results of empirical studies, but these cover a broad range of topics, including the use of computers in statistics instruction, teaching materials, and teaching strategies. A large portion of the nonempirical literature is devoted to descriptions of statistics courses and specific lessons that, though untested, still provide a resource for instruction. Recently numerous nonprint (electronic) resources for instruction, problem solving, and discussions about statistics instruction have also become available. These include many data sets and other instructional resources, statistics discussion groups, and the electronic Journal of Statistics Education.

Item Characteristics and Gender Differences on the SAT-M for Mathematically Able Youths

Item responses of a group of mathematically talented junior high school students were studied to explore the significant advantages that have been reported for males on the mathematical sections of the Scholastic Aptitude Test (SAT-M). Rasch latent trait item difficulties were estimated, and a regression-like linear analysis was used to examine the interactions of item features with sex of examinees. SAT-M items were classified by subject-matter content, item format, section membership, and whether a figure was presented with the item; their serial position also was considered. Examinees were grouped by sex, grade, and level of performance. Data sufficiency items were easier for all examinees than were multiple-choice items, and were much easier for the girls than for the bosys. Serial position of the items was important, with later items being more difficult, and girls found the first and last items “more extreme” than did the boys. Girls found miscellaneous items to be much easier than did the boys, but found the algebra items more difficult. Differences between mean item difficulties of boys and girls for arithmetic and geometry items were negligible. Results suggested that a number of specific aspects of mathematical performance were being measured by the SAT-M for this group of able youths. The patterns of response suggested weaknesses and strengths that might be a focus of mathematics instruction for these bright students.

An Effect Size for Regression Predictors in Meta-Analysis

A new effect size representing the predictive power of an independent variable from a multiple regression model is presented. The index, denoted as rsp, is the semipartial correlation of the predictor with the outcome of interest. This effect size can be computed when multiple predictor variables are included in the regression model and represents a partial effect size in the correlation family. The derivations presented in this article provide the effect size and its variance. Standard errors and confidence intervals can be computed for individual rsp values. Also, meta-analysis of the semipartial correlations can proceed in a similar fashion to typical meta-analyses, where weighted analyses can be used to explore heterogeneity and to estimate central tendency and variation in the effects. The authors provide an example from a meta-analysis of studies of the relationship of teacher verbal ability to school outcomes.